Maximum likelihood based analysis of equally spaced longitudinal count data with first-order antedependence and overdispersion

SpringerPlus

Table 1 Estimated parameters from the ML, GEE, and Poisson models in the analysis of the doctor visits data

Parameter	Estimate	SE	Wald	Pr(>\|W\|)
Coefficients
ML approach (\(AIC = 11707; BIC = 11{,}750\))
(Intercept)	−0.461	0.2811	2.69	0.1008
Reform	−0.113	0.0241	21.99	<0.0001
Age	0.005	0.0014	12.22	0.0005
Education	−0.008	0.0064	1.54	0.2153
Marital status	0.026	0.0294	0.75	0.3855
Health status	1.100	0.0313	1238.28	<0.0001
Log income	0.150	0.0376	15.83	<0.0001
Correlation parameters
Alpha	0.313	0.0208
GEE approach
(Intercept)	−0.381	0.5766	0.44	0.5083
Reform	−0.123	0.0529	5.40	0.0202
Age	0.005	0.0033	2.44	0.1182
Education	−0.009	0.0118	0.61	0.4349
Marital status	0.038	0.0698	0.30	0.5822
Health status	1.105	0.0873	160.23	<0.0001
Log income	0.139	0.0798	3.05	0.0809
Correlation parameters
Alpha	0.213	0.0238

Parameter	Estimate	SE	z value	Pr(>\|z\|)
Coefficients
Poisson regression (\(AIC = 11,899; BIC = 11,942\))
(Intercept)	−0.414	0.2691	−1.54	0.1242
Reform	−0.140	0.0265	−5.28	<0.0001
Age	0.004	0.0013	3.35	0.0008
Education	−0.011	0.0060	−1.78	0.0743
Marital status	0.041	0.0278	1.49	0.1375
Health status	1.133	0.0303	37.40	<0.0001
Log Income	0.149	0.0360	4.14	<0.0001